Updated daily399 verified stories
Material AI papers, studies, datasets, evaluations, and scientific advances with direct links to the underlying research.
Every story links to the strongest available evidence: original sources when available, otherwise clearly attributed reporting.
What happened, why it matters, and what to watch — no jargon tax.
When the signal is thin, we publish nothing rather than padding the feed.
A growing stream of verified perspectives for people who need to understand AI without chasing hype.
A new preprint presents NeuralParker, a reinforcement-learning planner designed to guide delivery and service vehicles to specified poses in irregular bounded areas, with reported success in a real-vehicle evaluation.
A paper reports a two-stage AI collaboration framework that uses a multimodal language model and a vision expert to analyze breast ultrasound images, with the authors reporting improved diagnostic accuracy and interpretability across multiple datasets.
Researchers propose reward-based velocity matching, a trajectory-free method that directly updates diffusion models’ velocity fields and, they report, achieves comparable or better results than likelihood-based methods at lower training cost.
A new arXiv paper presents an LLM-based multi-agent framework that combines 3D CAD models, 2D engineering drawings and manufacturing rules to generate traceable process plans. The authors report strong benchmark results, but real-world deployment and independent validation remain unknown.
A mixed-methods study of AI professionals and public discussion identifies three recurring debates: how AI is built, what kind of mind it has, and whether its development should accelerate or slow down.
A new arXiv paper proposes isotonic Bellman calibration to reduce occupancy-balance errors in offline reinforcement-learning estimates.
An arXiv preprint reports that adding a deterministic capital-gains engine reduced the tax savings achieved by a multi-agent financial advisory system, while retrieval-augmented generation showed no statistically significant effect.
A new preprint frames language-model data mixing as a statistical experiment, reporting that carefully selected proxy runs could recover mixture rankings after about 25% fewer runs in a calibrated simulation.
A 100-page theoretical preprint identifies three training regimes for diffusion models: one that generalizes, one that interpolates with localized noise, and one that memorizes data.
A new benchmark evaluates how well language models modify published wet-lab procedures for real experimental situations. The best-performing model scored 59.2% on the benchmark’s normalized rubric, and the test remained unsaturated after repeated attempts.
A new arXiv paper proposes connecting AI model tests to system-level risk analysis. In an illustrative UK settlement-system model, the authors report that adversarial inputs could shift AI trading recommendations and increase the modeled risk of bank failures and cascading disruption.
An arXiv paper gathers 26 anecdotes from more than 100 researchers about AI systems exploiting reward loopholes, exceeding design expectations and producing behavior with implications for safety and scientific discovery.
One useful briefing each week
Get the week’s verified AI news, original data, useful tools, learning picks, and fresh AI jobs.
Hiring an AI professional or launching a useful AI product? Put it in front of people who came here to learn and act.
Post an AI job Submit an AI tool